Structure

Article

The Structure of Dimeric Apolipoprotein A-IVand Its Mechanism of Self-AssociationXiaodi Deng,1 Jamie Morris,2 James Dressmen,2 Matthew R. Tubb,2 Patrick Tso,2 W. Gray Jerome,3W. Sean Davidson,2,*and Thomas B. Thompson1,*1Department of Molecular Genetics, Biochemistry, and Microbiology, College of Medicine, University of Cincinnati, OH 45267, USA2Department of Pathology and Laboratory Medicine, College of Medicine, Metabolic Diseases Institute, University of Cincinnati,Cincinnati, OH 45237, USA3Department of Pathology, Vanderbilt University Medical Center, Nashville, TN 37232, USA

*Correspondence: sean.davidson@uc.edu (W.S.D.), thompstb@ucmail.uc.edu (T.B.T.)

DOI 10.1016/j.str.2012.02.020

SUMMARY

Apolipoproteins are key structural elements of lipo-proteins and critical mediators of lipid metabolism.Their detergent-like properties allow them to emulsifylipid or exist in a soluble lipid-free form in variousstates of self-association. Unfortunately, these traitshave hampered high-resolution structural studiesneeded tounderstand thebiogenesisofcardioprotec-tive high-density lipoproteins (HDLs). We deriveda crystal structure of the core domain of human apoli-poprotein (apo)A-IV, an HDL component and impor-tant mediator of lipid absorption. The structure at2.4 A depicts two linearly connected 4-helix bundlesparticipating in a helix swapping arrangement thatoffers a clear explanation for how the protein self-associates as well as clues to the structure of itsmonomeric form. This also provides a logical basisfor antiparallel arrangements recently described forlipid-containing particles. Furthermore, we propose a‘‘swinging door’’ model for apoA-IV lipid association.

INTRODUCTION

Lipid transport in blood is accomplished by lipoproteins, nano-

meter-scale emulsions of lipid and protein that solve the funda-

mental ‘‘oil-in-water’’ solubility problem. One lipoprotein class,

high-density lipoproteins (HDLs), has been intensively studied

due to its association with various cardioprotective functions (re-

cently reviewed in Meurs et al., 2010). HDL proteins, called apoli-

poproteins, include (apo)A-I, apoA-II, apoA-IV, and many less

abundantspecies (Vaisaretal., 2007;Gordonetal., 2010b,2010a).

Many of these can interconvert between a soluble lipid-free state

and the lipid-bound form, with the lipid-free form able to interact

with the cell surface ATP binding cassette transporter (ABCA1)

to produce HDL particles (Rust et al., 1999; Lawn et al., 1999;

Brooks-Wilson et al., 1999; Remaley et al., 2001a). Unfortunately,

the details of this process are poorly understood, largely because

thebasicstructuresofboth the lipid-freeand -bound formsofmost

HDL apolipoproteins have not been clearly determined.

In terms of lipid-bound HDL apolipoproteins, much of our

structural understanding has come from the study of well-

defined recombinant HDL (rHDL) particles created in vitro

Structure 20

(reviewed in Davidson and Thompson, 2007). The heaviest focus

has been on apoA-I, which adopts a ‘‘double-belt’’ orientation in

which two apoA-I molecules wrap around a patch of phospho-

lipid bilayer in an antiparallel fashion (Koppaka et al., 1999;

Segrest et al., 2000). The overall tenets of this model have

been supported by numerous experimental studies, including

chemical crosslinking that confirmed the general inter-molecular

alignments of the resident apoA-I molecules (i.e., registry)

(Davidson and Hilliard, 2003; Bhat et al., 2005; Silva et al.,

2005b). Several refinements to the model have been proposed

that share the basic registry and antiparallel orientation (Bhat

et al., 2007; Martin et al., 2006; Wu et al., 2007). Despite this

understanding, fundamental questions remain as to how the

lipid-free forms transition from a structurally unknown lipid-free

state to this antiparallel arrangement in lipoprotein particles.

Unfortunately, progress on a high-resolution structural under-

standing of the lipid-free forms of HDL proteins has been slow.

Prior to last year, only three credible high-resolution structures

for human lipid-free apolipoproteins had been reported (Wilson

et al., 1991; Sivashanmugam and Wang, 2009; Borhani et al.,

1997). These have been helpful for understanding some aspects

of apolipoprotein function, though critical questions remain. The

difficulties in determining these structures can be attributed to

their tendency to form heterogeneous oligomeric arrays in the

lipid-free state. The functional role of this self-association, as

well as its structural basis, has been questioned for nearly

30 years. Recently, Mei and Atkinson reported the crystal struc-

ture of the N-terminal 184 residues of lipid-free human apoA-I

(Mei and Atkinson, 2011). The structure showed a half-circle

dimer with two helical bundles connected by shared antiparallel

helices. This ‘‘helix swapping’’ motif offered important structural

clues to the mechanism for apoA-I self association and also

a logical mechanism for lipoprotein particle formation. However,

the applicability of this motif and mechanism to other members

of the exchangeable apolipoprotein family is not known.

Human apolipoprotein (apo)A-IV is the largest apolipoprotein

(MW of 46 kDa). It is made in the gut and has been associated

with lipid absorption and chylomicron assembly (Stan et al.,

2003). As a prominent component of HDL, apoA-IV has also

been proposed to be a mediator of reverse cholesterol transport

(Emmanuel et al., 1994; Duverger et al., 1994; Main et al., 1996;

Remaley et al., 2001b), an antioxidant (Qin et al., 1998), a satiety

factor (Tso et al., 2001), a mediator of gut inflammation (Vowinkel

et al., 2004), and has recently been linked to b-amyloid clearance

in the brain (Cui et al., 2011). ApoA-IV shares several structural

, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved 767

Table 1. Data Collection, Phasing, and Refinement Statistics for MAD, SeMet

Nativea SeMeta

Data collection

Space group P6122

Cell dimensions

a, b, c (A) 70.8, 70.8, 512.4 71.2, 71.2, 513.4

a, b, g (�) 90, 90, 120 90, 90, 120

Peak (E1) Inflection (E2) Remote (E3)

Wavelength (A) 1.0332 0.9794 0.9796 0.9494

Resolution (A) 19.8-2.4 (2.53–2.4) 2.8 (2.9–2.8) 2.8 (2.9–2.8) 2.8 (3.0–2.8)

Rmerge 0.084 0.081 0.082 0.083

Mn (I/sd)b 19.4 (1.9) 16.5 (4.5) 16.4 (4.3) 17.3 (3.9)

Total reflection 485,811 220,658 220,938 242,804

Unique reflection 31,135 20,321 20,304 20,353

Completeness (%) 99.2 (99.3) 99.5 (99.8) 99.5 (99.8) 99.6 (99.8)

Redundancy 15.4 (15.9) 9.2 (9.5) 9.1 (9.5) 9.1 (9.5)

Refinement

Resolution (A) 18.8–2.4

Rwork (%)/Rfreec (%) 20.0/22.9

No. atoms 3,959

Protein 3,937

Water 22

B-factors

Protein 95.7

Water 93.5

Rmsd

Bond lengths (A) 0.010

Bond angles (�) 1.12

Ramachandran plot

Favored (%) 99.37

Outliers (%) 0.0

Two crystals were used for data collection, one native and one SeMet.aValues in parentheses are for highest-resolution shell as defined in the resolution row.bMn (I/sd) is defined as < merged < Ih > /sd(<Ih >) > � = signal/noise.cRfree is defined as 5% of initial total number of reflections. Rmsd bond lengths and angles are the deviations from ideal values.

Structure

Crystal Structure of Dimeric apoA-IV

characteristics with apoA-I and other exchangeable apolipopro-

teins, including putative amphipathic a-helical repeats that are

critical for lipid binding and self-association (Segrest et al.,

1994). Here, we describe the X-ray crystal structure of a stable

core domain of human apoA-IV. Although the apoA-IV structure

is consistent with the dimer arrangement observed in the

N-terminal apoA-I structure, including a similar helix-swapping

motif, significant structural differences are apparent in the

composition of the helical bundles. Analysis of the structure

suggests plausible molecular explanations for the structure of

monomeric apoA-IV and may provide important hints for how

apoA-IV may bind lipids.

RESULTS

X-ray Structure of Dimeric apoA-IV64-335

Numerous attempts to crystallize full-length apoA-IV have thus

far failed to provide diffraction quality crystals. However, we

768 Structure 20, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights

previously showed that lipid-free apoA-IV exhibits a compart-

mentalized structure with a large, well-folded central core and

less organized N and C termini that likely interact (Tubb et al.,

2007). Using limited proteolysis, we localized the core domain

to residues 64–335 (Tubb et al., 2008), and isolated dimers of

this modified protein formed diffraction quality crystals. The

crystal structure of this domain was determined to 2.4 A resolu-

tion by MAD with two molecules in the asymmetric unit (Table 1).

The model consisted of residues 74–312 but lacks interpretable

electron density for 10 and 23 residues at the N and C termini,

respectively (Figure 1A). The structure revealed a long rod-like

dimer (170 A L 3 20 A H 3 20 A W) consisting of two 4-helix

bundles stackedend-to-end in opposingorientations (Figure 1B).

There is a progressive right-handed twist that turns the structure

by 90� over the length of the rod. The helical hydrophobic faces

are directed toward the center of the bundles as seen in other

exchangeable apolipoproteins (Wilson et al., 1991; Wang

et al., 2002). Themonomers, each consisting of 4 helices (labeled

reserved

Figure 1. Structure of apoA-IV64-335

(A) Linear diagram of human apoA-IV. Arrows indicate the extent of the protein used for crystallization (64–335). The residues that were visible in the structure

(74–312) are shown with the four major helical domains (A–D) indicated as orange boxes.

(B) The structure, rotated through 90�, is depicted as blue and green ribbons interacting through a helical domain swap mechanism. The green monomer is also

shown in the absence of the blue monomer.

(C) Schematic representation of the dimer pulled apart.

(D) Space filling representation of the dimer with molecular dimensions indicated. A cavity at the center of the structure is highlighted, showing the exposure of

hydrophobic core residues in red (see also Figure S1).

See also Figure S8.

Structure

Crystal Structure of Dimeric apoA-IV

A–D), are related by 2-fold symmetry across an axis perpendic-

ular to the rod (Figure 1C). The N-terminal Helix-A (74–94) runs

perpendicular to the rod and caps the outer end of the helical

bundle. Helix-B (96–204) is extremely long, spanning the entire

length of the dimer. Following a 180� turn, Helix-C (206–255)

and Helix-D (265–312) form antiparallel helical arms of �80 A

in length and complete the other half of the bundle. The loops

connecting Helix-C and D (256–264, C-D loop) in each monomer

oppose each other at the center of the rod, causing a break in the

hydrophobic core. This results in a large cavity (>700 A3) (Dundas

et al., 2006) that is lined at the basin with exposed hydrophobic

residues from Helix-B of each monomer and topped with small

polar residues from the C-D loop (Figure 1D; Figure S1 available

online). The N and C termini from opposing chains are in prox-

imity at each end of the rod, consistent with an intermolecular

interaction that we proposed previously (Tubb et al., 2007).

Each 4-helix bundle in the dimer is composed of three helices

from one chain with the fourth stretching across from the

opposing chain. This interlocking interaction offers an immediate

explanation for apoA-IV dimerization with the interface burying

�5,900 A2 of predominately hydrophobic surface on eachmono-

mer. In fact, relative to the quaternary interfaces of other protein

assemblies, the ratio of buried surface area at the dimer interface

to total solvent accessible surface area is exceptionally high

(Ponstingl et al., 2005).

Structure ValidationSince crystallization typically occurs at high protein concentra-

tions and apolipoproteins tend to adopt interconverting oligo-

Structure 20

meric species, we set out to verify that the crystallization-state

matched the solution-state. In solution, apoA-IVWT exists as

a mixture of monomers and dimers (Pearson et al., 2004), as

does apoA-IV64-335 (Figure 2A). The helix sharing structure we

observed in apoA-IV64-335 and the large amount of hydrophobic

surface buried at the dimer interface suggest high dimer stability

with a correspondingly slow rate of dissociation. In fact, whenwe

purified the apoA-IV64-335 dimer by size exclusion chromatog-

raphy (SEC) and then reapplied the pooled dimer to the column,

little to no dissociation into monomer occurred at room temper-

ature or below (Figure S2A). This was verified by sedimentation

velocity (Figure S2C). These observations demonstrated that

the apoA-IV64-335 dimer is a stable species, thus making it

possible for us to characterize it in solution. Circular dichroism

(CD) spectroscopy of the isolated dimeric apoA-IV64-335 esti-

mated the helical content at 83% (Figure 2B). The visible crystal

structure of apoA-IV64-335 is about 87% helical. These values are

in remarkable agreement, especially if one assumes that the

invisible regions at the termini are largely random coil. Further-

more, lysine residues crosslinked in dimeric apoA-IV64-335 by

Bis(Sulfosuccinimidyl) suberate (BS3) (Table S1) were consistent

with the structure and well within the 23 A distance cutoff of BS3.

These data gave confidence that the visible crystal structure was

a highly plausible model for the solution structure of dimeric

apoA-IV64-335.

Satisfied that the structure model recapitulates soluble apoA-

IV64-335, we next evaluated how well apoA-IV64-335 served as

a structural surrogate for apoA-IVWT. We found that, despite

differences at the termini, the two proteins had remarkably

, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved 769

Figure 2. Ability of apoA-IV64-335 to Model

apoA-IVWT

(A) Isolated apoA-IVWT and apoA-IV64-335 dimer

incubated at 37�C in PBS for 24 hr and analyzed by

SEC (dimers at 11–12 ml, monomers at 14–15 ml).

(B) Far UV CD spectra of dimeric apoA-IVWT and

apoA-IV64-335 at 0.85mM at 25�C.(C) Thermal denaturation of both forms as moni-

tored by CD at 222 nm.

(D) Rate of dimer disassociation over time

measured by SEC for both apoA-IVWT and apoA-

IV64-335.

(E) Introduction of single Cys point mutations

(N151C, E263C, K233C, and R284C) to test the

dimer model. Predicted disulfide bond formation

in WT apoA-IV based on the crystal structure is

indicated with a solid line.

(F) SDS-PAGE of point mutations of apoA-IVWT

designed in (D) that were purified under reducing

conditions (right) and then oxidized to allow

disulfide bond formation (left), stained with

Coomassie blue.

See also Figure S2 and Table S1.

Structure

Crystal Structure of Dimeric apoA-IV

similar characteristics. For example, both versions exhibited

highly stable dimers that could be isolated by SEC and studied

(Figures S2B and S2C). They exhibited similar CD spectra and

thermal denaturation profiles with a Tm of 53�C ± 0.5 (Figures

2B and 2C), though the WT protein exhibited slightly less

coopertivity. In addition, at 37�C, the isolated dimers of both

forms dissociated with similar kinetics and reached a similar

770 Structure 20, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved

monomer/dimer ratio at equilibrium

(60% and 40%, respectively, Figure 2D).

This suggests that the majority of the

amino acids responsible for the stability

and self-association of apoA-IVWT are

present and functional in apoA-IV64-335.

Finally, Lys residues found to be cross-

linkable in apoA-IVWT (Tubb et al., 2008)

were highly plausible in the crystal struc-

ture of apoA-IV64-335. In total, these data

indicate that the apoA-IV64-335 crystal

structure is an appropriate template for

making structural predictions in the full-

length protein.

We further tested this by strategically

introducing Cys into apoA-IVWT (which

normally lacks them) and examining

whether a disulfide-bonded dimer could

form. Two ‘‘symmetrical’’ amino acids

coming into close contact in the center

of the dimer were identified that would

potentially form intermolecular disulfide

linkages in the apoA-IV64-335 dimer (Fig-

ure 2E). In the first mutant, apoA-

IVN151C, a Cys was introduced in the

center of Helix-B at a predicted point of

direct alignment across the dimer inter-

face. In apoA-IVE263C, the Cys was

placed in the C-D loop at a point of pre-

dicted alignment at the tips of the two helical arms. Figure 2F

shows that after oxidation both apoA-IVN151C and apoA-IVE263C

existed primarily as disulfide linked dimers in the lipid-free state

when analyzed by nonreducing PAGE, consistent with the

crystal structure (note: the apparent MW of the mutant dimers

is likely different because of gel migration effects related to the

location of the disulfide bond; Thomas et al., 2006). Under

Structure

Crystal Structure of Dimeric apoA-IV

reducing conditions, all mutants reverted tomonomers, confirm-

ing that the MW differences observed in the nonreduced PAGE

were due to the disulfides. In contrast, apoA-IVK233C and

apoA-IVR284C, where the introduced Cys are predicted to be 94

and 71 A apart in the crystal structure, remained mostly mono-

meric under oxidizing conditions (Figure 2F). Collectively, these

data indicate that the crystal structure is consistent with that of

soluble dimeric apoA-IV64-335, and most importantly, full-length

apoA-IV.

Distinct Segmentation of apoA-IVA well-recognized theme from sequence analyses of the

exchangeable apolipoproteins is the repeating motif of 22 aa

amphipathic a helices (Segrest et al., 1994). Consistent with

this paradigm, our structure shows that each monomer is com-

posed of amphipathic helical segments separated by ‘‘kinks’’

that are associated with helix-breaking residues, most com-

monly proline. Remarkably, the kinks align with the opposing

monomer along the shared Helix-B and within each monomer

between Helix C and D, despite their antiparallel orientation

(Figure 3A). These residues are highly conserved across species

(Figure S3), indicating a key structural and functional role. Six

pairs of proline residues align vertically and partition the B

helices into five even segments of 22 residues each (Figure 3A).

A set of paired proline residues (Pro 95 and Pro 205) caps the B

helices at either end. The helical arms composed of Helix-C and

D are also segmented, but differently than in Helix-B. Here two

prolines in the helical arm domain (249 in Helix-C, and 289 in

Helix-D) do not alignwith each other, but each alignswith a highly

conserved glycine or serine residue (Figure 3A; Figure S3). The

segmentation of Helix-C and D is flanked at one end by the

turn that connects the two helices, which consists of three

conserved glycine residues and Pro 311 of Helix-D at the other

end. This partitions the helical arm into three segments that

vary in length from 11–20 residues. Taken together, the core

domain of the apoA-IV64-335 dimer is divided into 11 paired

helical segments, five that are intermolecular and 6 that are intra-

molecular (Figure 3A).

Since prolines typically destabilize a helix extension, it was

surprising to find several distributed throughout the remarkably

long Helix B, inducing only minor directional kinks (Table S2).

Of the proline residues that do not cause a turn, the mean kink

angle was 16� (±4�), which is less than the average reported

value of 26� (Barlow and Thornton, 1988). Certainly crystal-

packing could be partly responsible for these observations, as

described in further detail (Deng et al., 2012), and in Figure S4.

Upon further examination of these kinks, their alignment across

the helical interface revealed an interesting pattern. While the

prolines were clearly associated with the kinks, they are offset

from each other by approximately half of a helical turn toward

the N terminus of each helix. The actual axis of each bend was

centered on a conserved alanine residue located two amino

acids C-terminal to each proline (Figure 3B). The intervening

residue was always a tyrosine or histidine, both large ringed resi-

dues, creating a P(Y/H)A structural motif. When two of these

motifs are aligned in an antiparallel fashion, as occurs throughout

the length of the shared B helices, the alanines come into direct

alignment. The side chains of the alanine residues point into

the bundle while those of the ringed residues are positioned on

Structure 20

the outside (Figures 3B and 3C; Figure S8). The disparity in

atomic volumes between the ringed residues and the alanines

indicates that further bending of the helix at the kink position

would only be favorable in one direction. Therefore, the aligned

P(Y/H)A motif appears to create a ‘‘door hinge’’ mechanism by

which the kink can only bend inward toward the hydrophobic

core. As bending progresses, the small side chains of the

alanines allow them to pinch together. Movement in the opposite

direction (i.e., bending away from the hydrophobic core) appears

unlikely due to steric hindrances and unfavorable energetics,

requiring a flip in the carbonyl geometry of each proline. The

P(Y/H)A motifs situated at the ends of the bundle (residues

95–97 and 205–207) provide examples of a fully bent hinge (Fig-

ure 3C). Remarkably, all the P(Y/H)A kinks in themolecule appear

designed to bend toward the hydrophobic core, irrespective of

their location. This directional hinging may be important for

structural transitions during lipid binding (see below).

Interhelical Interactions and Bundle StabilityDuring lipid binding, the helical bundles of apolipoproteins

must partially or completely transition from protein:protein inter-

actions to protein:lipid interactions. Therefore, we examined

features of the structure that offer insight into bundle stability,

such as hydrophobic interactions, hydrogen bonds and salt

bridges. Surprisingly, only a modest number of hydrogen bonds

were observed across the chains of the dimer as compared to

interfaces of other oligomeric assemblies (Figure S7). More

significant were the number of salt bridges, specifically at the

interface of Helix-B and C, which were clustered in two patches

and an intermolecular patch that links Helix-B to D (Figure 3D).

Although these patches form networks of hydrogen bonds/salt

bridges, their contribution to stabilizing the bundle may be less

than optimal. Due to the spatial arrangement of the charged

side chains, comparable stabilizing interactions can form across

helices as well as within the helix. This competition can be envi-

sioned to reduce the free energy contribution of these interac-

tions to the bundle stability (Figure 3D). Therefore, the hydrogen

bonds and salt bridges observed in the bundle appear more

favorable for aligning the helices than for stabilizing the bundle.

At the N terminus, the amphipathic Helix-A caps the ends of

the bundle and is involved in a number of interhelical interac-

tions. The hydrophobic face of Helix A wedges between

Helix-C and D of the opposing molecule and is anchored by

interactions coordinated through two conserved arginine resi-

dues (Arg 90 and 92) (Figure 3E; Figure S3). This splits the

ends of Helix-C and D, widening the bundle and creating a small

cavity in the hydrophobic core filled with ten leucines. The loose

packing of hydrophobic residues in this area could decrease the

overall stability of the bundle. These features may offer an expla-

nation for the overall low melting temperature and stability of

apoA-IV (Weinberg and Spector, 1985a).

Mechanism for Self-AssociationUnlike apoA-I and apoE, apoA-IV’s high oligomeric stability

afforded us the first opportunity to study the conformation of

an apolipoprotein monomer and dimer in isolation. Figure 4A

shows that the far UV CD spectra of the apoA-IVWT monomer

and dimer were essentially identical, indicating a similar degree

of secondary structure. Upon thermal denaturation, the two

, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved 771

Figure 3. Molecular Characteristics of the apoA-IV64-335 Dimer

(A) Alignment of helix disrupting ‘‘kinks’’ reveals paired segmentation. Proline, glycine, and serine residues are represented by spheres and colored accordingly.

(top) Helix-B from both green and blue chains run antiparallel and conserved proline residues from each helix are in a shifted alignment, dividing the B helices into

five paired segments of 22 aa in length. (bottom) Segmentation is also observed between Helix-C and D, where proline residues align vertically with glycine and

serine. Additional glycine residues that form the Helix-C to D turn are also displayed with the central glycine (Gly 260) annotated.

(B) Illustration of the P(H/Y)A motifs that align vertically across Helices-B looking down onto the two stacked helices (hydrophobic core at the top and polar

exterior on the bottom). Shown are residues PHA (161-163) and PYA (139-141) from opposing B helices. The potential axis of bending and direction is indicated

with arrows around dashed line.

(C) Illustration of paired P(H/Y)A motifs that align in turns (95–97 linking Helix-A to B) and (205–207 linking opposing Helix-B to C). Orientation is looking down the

length of the dimer rod.

(D) Hydrogen bonding and salt bridges at the interface of Helix-B and helical arms (Helix-C and D). (top) Intramolecular interactions are shown between acidic

(red) and basic (blue) residues looking down on the dimer rod parallel to the plane made up by the stacked B-helices. (bottom) Intermolecular interactions are

shown looking up in the same plane (i.e., rotated 180�).(E) Helix-A caps the 4-helix bundle at each opposing dimer end. Shown are interactions of two conserved arginine residues (90 and 92) within Helix-A that interact

at the top of the bundle. Interior leucine residues are shown as spheres and colored in accordance with each monomer.

See also Figures S3, S4, and S8, and Table S2.

Structure

Crystal Structure of Dimeric apoA-IV

772 Structure 20, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved

Structure

Crystal Structure of Dimeric apoA-IV

forms exhibited similar unfolding curves with equivalent Tmvalues of 53�C (Figure 4B). Finally, our crosslinking studies

showed only minor differences in the linkage pattern between

the two forms (Figure 4C). Taken together, these data strongly

indicate that the overall protein fold of the monomer and dimer

are highly related and make similar molecular contacts irrespec-

tive of the presence of one or two molecules.

Based on these observations and the domain swapping

apparent in our crystal structure, we envisioned a straightfor-

ward model for the monomeric form. Consider the hypothetical

separation of the dimer to produce an elongated monomer,

such as shown in Figure 4D. If a hinge is present near the middle

of the long Helix-B, the N-terminal half could bend back toward

the helical bundle and lie in the hydrophobic trough normally

occupied by the same helix from an opposing monomer (Fig-

ure 4H). This action can be likened to closing the extended

blade of a pocket knife. In this configuration, the molecular

contacts, as measured by crosslinking, will be similar to those

found in the dimer, except that those involving the N-terminal

half of Helix-B would be intramolecular instead of intermolecular

in the dimer. This should not dramatically change the helicity of

the monomer versus the dimer, consistent with the CD data, nor

should it affect the thermal denaturation properties of the two

forms on a molar basis. Thus, we reasoned that placing turn-

favoring residues in the middle of Helix-B might shift the oligo-

meric equilibrium in favor of the monomer by allowing Helix-B to

bend back instead of domain swapping with another apoA-IV

molecule. A candidate site for such a turn is at the kink

sequence following Pro161 (Figure 4D). A GOR4 secondary

structure prediction of this region showed a high propensity

for helix formation but, by substituting three glycines (GGG) in

place of HAD immediately after Pro 161, random coil is pre-

dicted between residues 158–164. This mutation was per-

formed in both apoA-IVWT and apoA-IV64-335. In sedimentation

velocity experiments, isolated monomers of both apoA-IVGGG

and apoA-IV64-335GGG exhibited MW consistent with their

respective monomer MW (Figure 4E; Figure S5). Isolated mono-

mers of both GGG mutants were thermally denatured, 2 min at

100�C, at various concentrations and then allowed to reanneal

for 10 min at room temperature. A native PAGE analysis (Fig-

ure 4F) showed that apoA-IVWT distributed between monomer

and dimer in a concentration-dependent manner, with dimer

favored at higher concentrations (Note: on native gels, apoA-

IV tends to run larger than predicted by its MW, possibly due

to its extended conformation). By contrast, apoA-IVGGG

completely failed to form dimers even at the highest concentra-

tion. The results were identical for apoA- IV64-335 and its corre-

sponding GGG mutant. The same conclusions were obtained

when SEC, rather than native PAGE, was used as the readout

for extent of dimerization (Figure 4G). These results strongly

argue that introduction of a GGG ‘‘hinge’’ shifts the oligomeriza-

tion equilibrium of apoA-IV toward the monomer, consistent

with the structural scheme shown in Figure 4H.

Particle FormationMost apolipoproteins can form rHDL via a number of mecha-

nisms, including spontaneous association with short chain

phospholipids such as dimyrstoyl phosphatidylcholine (DMPC).

Using the conditions in Experimental Procedures, apoA-IVWT

Structure 20

produced DMPC-containing particles with a hydrodynamic

diameter of 140 A as measured by native PAGE. ApoA-IV64-335

also produced an rHDL particle, though slightly smaller at about

132 A (Figures 5A and 5B; Figure S6). The chemical composi-

tions of both particles are shown in Table S3. Furthermore, EM

images (Figure 5A) showed that these particles are disc-shaped

and can stack into characteristic structures called rouleux,

a well-noted behavior for apoA-I and other exchangeable

apolipoproteins (Nichols et al., 1983). The particle diameters as

determined by EM were larger than most estimated by native

PAGE for both apoA-IVWT and apoA-IV64-335 (16 and 15.5 nm,

respectively), but both particles were similar in size and

morphology. This suggests that the core four-helical bundle

domain identified in the structure of apoA-IV64-335 exhibits similar

particle forming characteristics as apoA-IVWT, including overall

particle size.

DISCUSSION

Compared to fully soluble proteins, only a handful of apolipopro-

tein structures have been solved by X-ray crystallography

or nuclear magnetic resonance. We believe that the current

structure offers significant new insight into long held questions

concerning self-association and lipoprotein assembly that may

be applicable to the larger family of exchangeable apolipopro-

teins and their role in HDL biogenesis.

Mechanism for Self-AssociationThe physiological role and the mechanism for apolipoprotein

self-association have long been questioned, but the molecular

basis for the interactions have only just started to be resolved

with the recent crystal structure of lipid-free apoA-I (Mei et al.,

2011). Both dimeric apoA-I and apoA-IV are bona fide examples

of significant 3D domain swaps that provide a clear mechanism

for how the proteins self-associate. This departs substantially

from previously proposed mechanisms that invoked interaction

between charged or hydrophobic patches between indepen-

dently folded monomers (Silva et al., 2005a; Weinberg and

Spector, 1985b). In the case of apoA-IV, the monomer and dimer

forms both exist in stable low energy states that interconvert

slowly enough so that they can be studied in isolation. We

found that both the dimer and monomer exhibit highly similar

structures as determined by CD and chemical crosslinking.

Indeed, our introduction of a flexible region in the center of

the long, swapped helix resulted in a form that remained

monomeric (Figure 4H). This provides the first detailed picture

of an apolipoprotein monomer/dimer relationship, but its appli-

cability to other apolipoproteins, including apoA-I, remains to

be determined.

Lipoprotein Particle BiogenesisAs stated in the Introduction, the most well supported model for

the structure of nascent HDL is the double belt model in which

two apolipoprotein molecules (typically apoA-I) are arranged in

a ring-like, antiparallel structure that encapsulates a lipid bilayer

(Li et al., 2004). A basic question in lipoprotein biology relates to

the mechanism by which lipid-free proteins, presumably starting

out as four-helical bundles, interact with each other and lipid to

generate such an antiparallel structure. A key feature of these

, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved 773

Figure 4. Transition of apoA-IV Dimer to Monomer

(A) Far UV circular dichroism (CD) spectra of the SEC isolated dimer and monomer forms.

(B) Thermal denaturation of isolated dimer and monomer as monitored by CD.

(C) A plot of crosslinking patterns derived from the isolated apoA-IVWT monomer (upper left, filled symbols) and dimer (lower right, open symbols). The X and

Y-axes show the sequence number of the Lys residue involved in a specific crosslink (both intra- and intermolecular). Crosslinks indicated with a filled triangle

were unique to the monomer. The high similarity of the crosslinking patterns on both sides of the unity line dividing the plot is clear, indicating overall similarity in

the tertiary folds of the two forms.

Structure

Crystal Structure of Dimeric apoA-IV

774 Structure 20, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved

Figure 5. Particle Formation Comparison of apoA-IVWT and apoA-IV64-335

(A) Negative stain electron microscopy images of reconstituted lipoprotein DMPC particles generated with apoA-IVWT (top) and apoA-IV64-335 (bottom), with their

respective particle diameter measurement distribution.

(B) Native PAGE gel of reconstituted lipoprotein DMPC particles generated with apoA-IVWT and apoA-IV64-335, Coomassie stained.

See also Figure S6 and Table S3.

Structure

Crystal Structure of Dimeric apoA-IV

structures, apoA-I1-184 and apoA-IV64-335, is that their dimers

already have significant portions of the participating molecules

overlapping in an antiparallel manner. Combining this observa-

tion with the concept of unidirectional hinges mediated by the

segmental kinks described above, we propose that an apoA-IV

rod-like dimer could form a discoidal lipoprotein particle

with only minor adjustments of these hinge angles (Figure 6).

Beginning with the dimer rod, one can imagine inserting phos-

pholipid molecules into the hydrophobic space located between

the B helices and the helical arms composed of Helix-C and D.

As lipids are added, the kink/hinges bend toward the core,

allowing the helical arms or ‘‘doors’’ to swing away from the

Helix-B backbone. Continued lipid accumulation results in

a circular ring with the overall curvature mediated by the

segmental kinks. One potential result is a ring with about half

the circumference composed of the antiparallel Helix-B’s,

making intermolecular contacts (Figures 6B and 6C). The other

half is composed of the two antiparallel helical arms making

intramolecular contacts. Alternatively, the helical arms may

(D) Placement of the glycine hinge introduced into Helix-B. Depicted is one compo

core in orange. Introduced glycines are shown in yellow. The arrow indicates ho

residues to form a monomer.

(E) Sedimentation velocity molecular weight distribution plot of monomer ap

(30.6 ± 3.9 kDa).

(F) Native gels showing the redistribution of monomer and dimer of apoA-IV varia

Concentrations of protein are indicated for each lane.

(G) Thermally denatured then re-annealed samples at 1.5 mg/ml from (F) analyze

(H) Proposed model of transition from dimer to monomer where, in the absence

a monomolecular 4-helix bundle.

See also Figure S5.

Structure 20

open and exchange their C termini, resulting in a closed ring

that is entirely stabilized by intermolecular interactions as in

the apoA-I double belt (Figures 6D and 6E) and the scheme

proposed by Mei and Atkinson. Our preliminary modeling

analysis indicates that both scenarios could easily generate

discoidal particles that are consistent with the particle diameters

and compositions that we measured experimentally. The model

does not define how phospholipid enters the hydrophobic

space, however, one can speculate that lipid can access the

core through the central cavity or the loose bundle packing

created by Helix-A that caps the 4-helix bundles. Our crystal

structure should serve as an excellent guide for further mutagen-

esis experiments designed to sort out this issue and will be valu-

able for future studies designed to understand the structure of

lipid-bound apoA-IV.

Comparison to apoA-IAs indicated above, the dimeric structure of apoA-IV65-335

exhibited several striking similarities to the recent structure of

nent of the dimer with the polar surface in green and the exposed hydrophobic

w Helix-B might fold back onto the bundle to bury the exposed hydrophobic

oA-IVWT (40.4 ± 4 kDa), apoA-IVGGG (39.8 ± 3 kDa) and apoA-IV64-335GGG

nts that were subjected to thermal denaturation and subsequently reannealed.

d by SEC.

of a second molecule, the extended Helix-B folds back on itself, completing

, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved 775

Figure 6. Hypothesized Model for apoA-IV rHDL

(A) A segmented apoA-IV64-335 dimer is shown. One chain

is colored rainbow from N to C, where the other is colored

in gray scale with (circle) being the N and (diamond) being

the C terminus. The arrows indicate a proposed outward

swing of the helical arm domains as lipid accumulates in

the hydrophobic core.

(B–E) Two models are proposed that either keep the arms

intact after they swing open (B and C) or the arms swap

helices (D and E).

See also Figure S7.

Structure

Crystal Structure of Dimeric apoA-IV

apoA-I1-184 (Figure 7). First, residues 79–182 of apoA-I formed

a long helix that participated in an antiparallel domain swapping

interaction similar to apoA-IV Helix-B. Unlike the relatively

straight apoA-IV structure, apoA-I was bent into a semicircular

shape owing to more extensive kink angles between helical

domains of about 25� versus 16� for apoA-IV (Table S2), nearer

the average reported value of 26� (Barlow and Thornton, 1988).

Like apoA-IV, these kinks were associated with conserved

Pro residues, though the P(Y/H)A motif prevalent in apoA-IV

was not recapitulated in apoA-I. This provides strong evidence

that domain swapping may be a common motif among

exchangeable apolipoproteins that may mediate apolipoprotein

self-association. Second, the apoA-I structure also features

helix ‘‘arms’’ that fold across the antiparallel backbone. Interest-

ingly, these are formed by the N-terminal 65 aa in apoA-I, but

are composed of elements near the C-terminal end of the

apoA-IV (206–312). Due to this difference, the transition from

the helical arm to the paired helix is distinct for apoA-I and

apoA-IV (Figure 7C), resulting in different capping mechanisms

of the 4-helix bundle (Figure 7D). In apoA-I1-184, the helix arm

transitions into the antiparallel helix via a ten residue helix

(67–78), which caps the 4-helix bundle. By contrast, in apoA-

IV64-335 the paired helix transitions into the helical arm via

a proline-induced turn (205–207), with the bundle being capped

by Helix-A of the opposing molecule. Because of this difference

in architecture, if a similar ‘‘swinging door’’ model is proposed

for apoA-I, the helical arms become crossed in relation to the

long paired helix, requiring a crossover event in the double-

belt model. Though the arrangement of the two structures is

slightly different, it is surprising that varying regions of the

polypeptide should give rise to apparently similar structural

features, especially given the view that the apoA-IV gene

evolved from a primordial form of apoA-I (Karathanasis et al.,

1986). This could be explained by a localized gene reversion

event during apoA-IV evolution. Beyond the presence of the

turns and amphipathic helices, we have not yet found common

architectural features that predispose these residues, either in

apoA-I or apoA-IV, to specifically form helical arms. However,

776 Structure 20, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved

we would point out that the missing sequences

in both apoA-IV and apoA-I may interact with

the bundle domains in ways that are not yet

clear. Thus, the propensity for helical arm

formation might not be exclusively encoded in

the arms themselves. Third, the arms in

apoA-I1-184 are significantly shorter, resulting

in a less defined 4-helix bundle motif com-

pared to apoA-IV64-335. As a consequence, a much larger cavity

is created in the center of the rod, reducing the stability of the

apoA-I1-184 helix 5 region. The authors propose that this flexible

region might be a site for interaction with lecithin:cholesterol

acyl transferase (LCAT). Our analysis of the corresponding

region in apoA-IV64-335 suggests that there is no noticeable

increase in B factor in this area with respect to other stretches

of helix. Furthermore, our demonstration that apoA-IVN151C

readily forms covalent dimers (Figure 2D) suggests that the

helices in this region remain in close contact. This potential

difference in conformational flexibility at the geometric center

of each molecule may explain the remarkable difference in

monomer/dimer dynamics between apoA-IV (slow) and apoA-I

(fast) as well as why apoA-I is the superior activator of LCAT

(Jonas et al., 1989). The availability of both of these struc-

tures will prove highly valuable for driving future studies

aimed at understanding the structure/functional relationships

of these and other exchangeable apolipoproteins associated

with HDL. This comes at a critical time when there is a major

effort within the pharmaceutical industry to develop thera-

pies that raise HDL with the goal of enhancing protection

from CVD.

EXPERIMENTAL PROCEDURES

Protein Expression and Purification

ApoA-IVWT, apoA-IV64-335, apoA-IVGGG, and apoA-IV64-335GGG were ex-

pressed and purified as described previously (Deng et al., 2012). Single Cys

mutants of apoA-IV were immediately subjected to 5 mM DTT and kept under

reducing conditions throughout the purification process. DTT was removed by

dialysis (48 hr at 4�C) to allow disulfide bond formation.

X-ray Structure Determination of apoA-IV64-335

The details of crystallization and structure determination have been previously

published (Deng et al., 2012). Diffraction data were collected at the Advanced

Photon Source (23ID-B GM/CA) at Argonne National Laboratory. Data on

the Se-Met substituted crystals were collected at three wavelengths and

utilized for MAD phasing (Table S2). The final structure has a MolProbity

(Chen et al., 2010) score of 1.79 (97th percentile), with no Ramachandran

outliers.

Figure 7. Structural Comparison of apoA-I1-184 and apoA-IV64-335

(A and B) Ribbon (top) and schematic (bottom) representation of apoA-I1-184(A) and apoA-IV64-335(B) in their dimeric forms, each with one chain colored in rainbow

from N to C terminus. Both structures reveal two long paired helices with anti-parallel orientation spanning similar distances and punctuated by proline residues.

The cross chain aligned proline residues (represented as P) divide the long anti-parallel helix into two 18 and three 22 aa paired helix regions for apoA-I1-182 and

five 22 aa paired helix regions for apoA-IV64-335. The length of the helical arms is shown, with the central cavity depicted as a yellow oval.

(C) Schematic representation of both structures linearly aligned at the paired helix segment. Note the flipped orientation of the helix arm region in relation to the

paired helix segment.

(D) Schematic view looking down the bundle showing different linkage from the helical arms to the paired helices and how the 4-helix bundle is capped in each

structure.

Structure

Crystal Structure of Dimeric apoA-IV

Crosslinking and Mass Spectrometry

Crosslinking with Bis(Sulfosuccinimidyl) suberate and subsequent analysis

by electrospray mass spectrometry was performed exactly as we have

described previously (Tubb et al., 2008). See Supplemental Experimental

Procedures for more detail.

Dissociation of Dimer

Isolated dimers of apoA-IVWT and apoA-IV64-335 were diluted to 0.15 mg/ml in

PBS and incubated at 37�C. Proteins were analyzed by SEC on a Superdex-

200 column at 4�C. Monomer and dimer peaks were integrated to determine

their ratio.

Circular Dichroism

Immediately preceding the circular dichroism (CD_ measurements, the

samples were analyzed by SEC to verify the oligomeric state. Far-UV wave-

length measurements were performed at 25�C with 1 nm steps from 185 to

300 nm using samples in 20 mM NaPO4, 50 mM NaF (pH 7.5). Data were

analyzed using the SELCON 3 method. Thermal denaturation experiments

were performed by monitoring the ellipticity at 222 nm from 15�C to 95�C in

1�C increments with 2 m temperature equilibrations and 5 s averaging time.

Structure 20

The measurements were performed in a 1 cm cuvette under constant stirring

with a protein concentration of 0.37 mg/ml.

Apolipoprotein-Lipid Particle Formation

DMPC (Avanti Polar Lipids, Birmingham, AL) in chloroform was dried down

under nitrogen, resuspended in PBS (pH 7.3), bath sonicated for 30 s, and

combined with apoA-IV at a 350:1 DMPC: apoA-IV molar ratio to a final

protein concentration of 0.8 mg/ml. The lipid:protein mixture was incubated

for 24 hr at 24.5�C followed by a 16 hr 37�C incubation. Particle homogeneity

was verified by native PAGE. Composition was determined by analysis of

phosphorus (Sokoloff and Rothblat, 1974) and modified Lowry assay (Mark-

well et al., 1978) from multiple independent rHDL preparations. Electron

microscopy of rHDL particles was performed as described previously (Huang

et al., 2011).

ACCESSION NUMBERS

The coordinates and structure factors of Homo sapiens apolipoprotein A-IV

have been deposited in the Protein Data Bank under the accession code

3S84.

, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved 777

Structure

Crystal Structure of Dimeric apoA-IV

SUPPLEMENTAL INFORMATION

Supplemental Information includes eight figures, three tables, and Supple-

mental Experimental Procedures and can be found with this article online at

doi:10.1016/j.str.2012.02.020.

ACKNOWLEDGMENTS

This work was supported by US National Institutes of Health (NIH) R01 grant

HL67093 (to W.S.D.), an NIH training grant T32 HL007382 (to X.D.), an

American Heart Association Pre-doctoral fellowship to M.R.T. and NIH R01

grant HL48148 (to W.G.J.). EM was carried out in the Vanderbilt University

Research Electron Microscopy Resource of the Cell Imaging Core (partially

supported by NIH grants CA68485, DK20593, and DK58404).

Received: November 27, 2011

Revised: February 2, 2012

Accepted: February 24, 2012

Published: May 8, 2012

REFERENCES

Barlow, D.J., and Thornton, J.M. (1988). Helix geometry in proteins. J. Mol.

Biol. 201, 601–619.

Bhat, S., Sorci-Thomas, M.G., Alexander, E.T., Samuel, M.P., and Thomas,

M.J. (2005). Intermolecular contact between globular N-terminal fold and

C-terminal domain of ApoA-I stabilizes its lipid-bound conformation: studies

employing chemical cross-linking and mass spectrometry. J. Biol. Chem.

280, 33015–33025.

Bhat, S., Sorci-Thomas, M.G., Tuladhar, R., Samuel, M.P., and Thomas, M.J.

(2007). Conformational adaptation of apolipoprotein A-I to discretely sized

phospholipid complexes. Biochemistry 46, 7811–7821.

Borhani, D.W., Rogers, D.P., Engler, J.A., and Brouillette, C.G. (1997). Crystal

structure of truncated human apolipoprotein A-I suggests a lipid-bound

conformation. Proc. Natl. Acad. Sci. USA 94, 12291–12296.

Brooks-Wilson, A., Marcil, M., Clee, S.M., Zhang, L.H., Roomp, K., van Dam,

M., Yu, L., Brewer, C., Collins, J.A., Molhuizen, H.O., et al. (1999). Mutations in

ABC1 in Tangier disease and familial high-density lipoprotein deficiency. Nat.

Genet. 22, 336–345.

Chen, V.B., Arendall, W.B., 3rd, Headd, J.J., Keedy, D.A., Immormino, R.M.,

Kapral, G.J., Murray, L.W., Richardson, J.S., and Richardson, D.C. (2010).

MolProbity: all-atom structure validation for macromolecular crystallography.

Acta Crystallogr. D Biol. Crystallogr. 66, 12–21.

Cui, Y., Huang, M., He, Y., Zhang, S., and Luo, Y. (2011). Genetic ablation of

apolipoprotein A-IV accelerates Alzheimer’s disease pathogenesis in a mouse

model. Am. J. Pathol. 178, 1298–1308.

Davidson, W.S., and Hilliard, G.M. (2003). The spatial organization of apolipo-

protein A-I on the edge of discoidal high density lipoprotein particles: a mass

specrometry study. J. Biol. Chem. 278, 27199–27207.

Davidson, W.S., and Thompson, T.B. (2007). The structure of apolipoprotein

A-I in high density lipoproteins. J. Biol. Chem. 282, 22249–22253.

Deng, X., Davidson, W.S., and Thompson, T.B. (2012). Improving the diffrac-

tion of apoA-IV crystals through extreme dehydration. Acta Crystallogr.

Sect. F Struct. Biol. Cryst. Commun. 68, 105–110.

Dundas, J., Ouyang, Z., Tseng, J., Binkowski, A., Turpaz, Y., and Liang, J.

(2006). CASTp: computed atlas of surface topography of proteins with struc-

tural and topographical mapping of functionally annotated residues. Nucleic

Acids Res. 34 (Web Server issue), W116–W118.

Duverger, N., Ghalim, N., Theret, N., Fruchart, J.C., and Castro, G. (1994).

Lipoproteins containing apolipoprotein A-IV: composition and relation to

cholesterol esterification. Biochim. Biophys. Acta 1211, 23–28.

Emmanuel, F., Steinmetz, A., Rosseneu, M., Brasseur, R., Gosselet, N.,

Attenot, F., Cuine, S., Seguret, S., Latta, M., Fruchart, J.C., et al. (1994).

Identification of specific amphipathic a-helical sequence of human apolipo-

778 Structure 20, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights

protein A-IV involved in lecithin:cholesterol acyltransferase activation.

J. Biol. Chem. 269, 29883–29890.

Gordon, S., Durairaj, A., Lu, J.L., and Davidson, W.S. (2010a). High-density

lipoprotein proteomics: Identifying new drug targets and biomarkers by under-

standing functionality. Curr Cardiovasc Risk Rep 4, 1–8.

Gordon, S.M., Deng, J., Lu, L.J., and Davidson, W.S. (2010b). Proteomic

characterization of human plasma high density lipoprotein fractionated by

gel filtration chromatography. J. Proteome Res. 9, 5239–5249.

Huang, R., Silva, R.A., Jerome, W.G., Kontush, A., Chapman, M.J., Curtiss,

L.K., Hodges, T.J., and Davidson, W.S. (2011). Apolipoprotein A-I structural

organization in high-density lipoproteins isolated from human plasma. Nat.

Struct. Mol. Biol. 18, 416–422.

Jonas, A., Kezdy, K.E., and Wald, J.H. (1989). Defined apolipoprotein A-I

conformations in reconstituted high density lipoprotein discs. J. Biol. Chem.

264, 4818–4824.

Karathanasis, S.K., Oettgen, P., Haddad, I.A., and Antonarakis, S.E. (1986).

Structure, evolution, and polymorphisms of the human apolipoprotein A4

gene (APOA4). Proc. Natl. Acad. Sci. USA 83, 8457–8461.

Koppaka, V., Silvestro, L., Engler, J.A., Brouillette, C.G., and Axelsen, P.H.

(1999). The structure of human lipoprotein A-I. Evidence for the ‘‘belt’’ model.

J. Biol. Chem. 274, 14541–14544.

Lawn, R.M., Wade, D.P., Garvin, M.R., Wang, X., Schwartz, K., Porter, J.G.,

Seilhamer, J.J., Vaughan, A.M., and Oram, J.F. (1999). The Tangier disease

gene product ABC1 controls the cellular apolipoprotein-mediated lipid

removal pathway. J. Clin. Invest. 104, R25–R31.

Li, L., Chen, J., Mishra, V.K., Kurtz, J.A., Cao, D., Klon, A.E., Harvey, S.C.,

Anantharamaiah, G.M., and Segrest, J.P. (2004). Double belt structure of

discoidal high density lipoproteins: molecular basis for size heterogeneity.

J. Mol. Biol. 343, 1293–1311.

Main, L.A., Ohnishi, T., and Yokoyama, S. (1996). Activation of human plasma

cholesteryl ester transfer protein by human apolipoprotein A-IV. Biochim.

Biophys. Acta 1300, 17–24.

Markwell, M.A., Haas, S.M., Bieber, L.L., and Tolbert, N.E. (1978). A modifica-

tion of the Lowry procedure to simplify protein determination in membrane and

lipoprotein samples. Anal. Biochem. 87, 206–210.

Martin, D.D., Budamagunta, M.S., Ryan, R.O., Voss, J.C., and Oda, M.N.

(2006). Apolipoprotein A-I assumes a ‘‘looped belt’’ conformation on reconsti-

tuted high density lipoprotein. J. Biol. Chem. 281, 20418–20426.

Mei, X., and Atkinson, D. (2011). Crystal structure of C-terminal truncated

apolipoprotein A-I reveals the assembly of high density lipoprotein (HDL) by

dimerization. J. Biol. Chem. 286, 38570–38582.

Meurs, I., Van Eck, M., and Van Berkel, T.J. (2010). High-density lipoprotein:

key molecule in cholesterol efflux and the prevention of atherosclerosis.

Curr. Pharm. Des. 16, 1445–1467.

Nichols, A.V., Gong, E.L., Blanche, P.J., and Forte, T.M. (1983).

Characterization of discoidal complexes of phosphatidylcholine, apolipopro-

tein A-I and cholesterol by gradient gel electrophoresis. Biochim. Biophys.

Acta 750, 353–364.

Pearson, K., Saito, H., Woods, S.C., Lund-Katz, S., Tso, P., Phillips, M.C., and

Davidson, W.S. (2004). Structure of human apolipoprotein A-IV: a distinct

domain architecture among exchangeable apolipoproteins with potential

functional implications. Biochemistry 43, 10719–10729.

Ponstingl, H., Kabir, T., Gorse, D., and Thorton, J.M. (2005). Morphological

aspects of oligomeric protein structures. Prog. Biophys. Mol. Biol. 89, 9–35.

Qin, X., Swertfeger, D.K., Zheng, S., Hui, D.Y., and Tso, P. (1998).

Apolipoprotein AIV: a potent endogenous inhibitor of lipid oxidation. Am. J.

Physiol. 274, H1836–H1840.

Remaley, A.T., Stonik, J.A., Demosky, S.J., Neufeld, E.B., Bocharov, A.V.,

Vishnyakova, T.G., Eggerman, T.L., Patterson, A.P., Duverger, N.J.,

Santamarina-Fojo, S., and Brewer, H.B., Jr. (2001a). Apolipoprotein specificity

for lipid efflux by the human ABCAI transporter. Biochem. Biophys. Res.

Commun. 280, 818–823.

Remaley, A.T., Stonik, J.A., Demosky, S.J., Neufeld, E.B., Bocharov, A.V.,

Vishnyakova, T.G., Eggerman, T.L., Patterson, A.P., Duverger, N.J.,

reserved

Structure

Crystal Structure of Dimeric apoA-IV

Santamarina-Fojo, S., and Brewer, H.B., Jr. (2001b). Apolipoprotein specificity

for lipid efflux by the human ABCAI transporter. Biochem. Biophys. Res.

Commun. 280, 818–823.

Rust, S., Rosier, M., Funke, H., Real, J., Amoura, Z., Piette, J.C., Deleuze, J.F.,

Brewer, H.B., Duverger, N., Denefle, P., and Assmann, G. (1999). Tangier

disease is caused by mutations in the gene encoding ATP-binding cassette

transporter 1. Nat. Genet. 22, 352–355.

Segrest, J.P., Garber, D.W., Brouillette, C.G., Harvey, S.C., and

Anantharamaiah, G.M. (1994). The amphipathic alpha helix: a multifunctional

structural motif in plasma apolipoproteins. Adv. Protein Chem. 45, 303–369.

Segrest, J.P., Harvey, S.C., and Zannis, V. (2000). Detailed molecular model of

apolipoprotein A-I on the surface of high-density lipoproteins and its functional

implications. Trends Cardiovasc. Med. 10, 246–252.

Silva, R.A., Hilliard, G.M., Fang, J., Macha, S., and Davidson, W.S. (2005a). A

three-dimensional molecular model of lipid-free apolipoprotein A-I determined

by cross-linking/mass spectrometry and sequence threading. Biochemistry

44, 2759–2769.

Silva, R.A., Hilliard, G.M., Li, L., Segrest, J.P., and Davidson, W.S. (2005b). A

mass spectrometric determination of the conformation of dimeric apolipopro-

tein A-I in discoidal high density lipoproteins. Biochemistry 44, 8600–8607.

Sivashanmugam, A., and Wang, J. (2009). A unified scheme for initiation and

conformational adaptation of human apolipoprotein E N-terminal domain

upon lipoprotein binding and for receptor binding activity. J. Biol. Chem.

284, 14657–14666.

Sokoloff, L., and Rothblat, G.H. (1974). Sterol to phospholipid molar ratios of

L cells with qualitative and quantitative variations of cellular sterol. Proc.

Soc. Exp. Biol. Med. 146 (Sep), 1166–1172.

Stan, S., Delvin, E., Lambert, M., Seidman, E., and Levy, E. (2003). Apo A-IV: an

update on regulation and physiologic functions. Biochim. Biophys. Acta 1631,

177–187.

Thomas, M.J., Bhat, S., and Sorci-Thomas, M.G. (2006). The use of chemical

cross-linking and mass spectrometry to elucidate the tertiary conformation of

lipid-bound apolipoprotein A-I. Curr. Opin. Lipidol. 17, 214–220.

Structure 20

Tso, P., Liu, M., Kalogeris, T.J., and Thomson, A.B. (2001). The role of apolipo-

protein A-IV in the regulation of food intake. Annu. Rev. Nutr. 21, 231–254.

Tubb, M.R., Silva, R.A.G.D., Pearson, K.J., Tso, P., Liu, M., and Davidson,

W.S. (2007). Modulation of apolipoprotein A-IV lipid binding by an interaction

between the N and C termini. J. Biol. Chem. 282, 28385–28394.

Tubb, M.R., Silva, R.A., Fang, J., Tso, P., and Davidson, W.S. (2008). A three-

dimensional homology model of lipid-free apolipoprotein A-IV using cross-

linking and mass spectrometry. J. Biol. Chem. 283, 17314–17323.

Vaisar, T., Pennathur, S., Green, P.S., Gharib, S.A., Hoofnagle, A.N., Cheung,

M.C., Byun, J., Vuletic, S., Kassim, S., Singh, P., et al. (2007). Shotgun proteo-

mics implicates protease inhibition and complement activation in the antiin-

flammatory properties of HDL. J. Clin. Invest. 117, 746–756.

Vowinkel, T., Mori, M., Krieglstein, C.F., Russell, J., Saijo, F., Bharwani, S.,

Turnage, R.H., Davidson, W.S., Tso, P., Granger, D.N., and Kalogeris, T.J.

(2004). Apolipoprotein A-IV inhibits experimental colitis. J. Clin. Invest. 114,

260–269.

Wang, J., Sykes, B.D., and Ryan, R.O. (2002). Structural basis for the confor-

mational adaptability of apolipophorin III, a helix-bundle exchangeable apoli-

poprotein. Proc. Natl. Acad. Sci. USA 99, 1188–1193.

Weinberg, R.B., and Spector, M.S. (1985a). Structural properties and lipid

binding of human apolipoprotein A-IV. J. Biol. Chem. 260, 4914–4921.

Weinberg, R.B., and Spector, M.S. (1985b). The self-association of human

apolipoprotein A-IV. Evidence for an in vivo circulating dimeric form. J. Biol.

Chem. 260, 14279–14286.

Wilson, C., Wardell, M.R., Weisgraber, K.H., Mahley, R.W., and Agard, D.A.

(1991). Three-dimensional structure of the LDL receptor-binding domain of

human apolipoprotein E. Science 252, 1817–1822.

Wu, Z., Wagner, M.A., Zheng, L., Parks, J.S., Shy, J.M., 3rd, Smith, J.D.,

Gogonea, V., and Hazen, S.L. (2007). The refined structure of nascent HDL

reveals a key functional domain for particle maturation and dysfunction. Nat.

Struct. Mol. Biol. 14, 861–868.

, 767–779, May 9, 2012 ª2012 Elsevier Ltd All rights reserved 779